I’m Salman Masih, a Ph.D. candidate in Computer Science, specializing in Natural Language Processing (NLP), with a focus on multilingual models, legal text summarization, and fairness in large language models for low-resource languages. My research aims to tackle challenges in applying advanced NLP techniques to legal texts in languages with limited annotated resources, developing models that are both efficient and accurateMy thesis explores abstractive summarization for low-resource languages, focusing on four key papers that examine transformer-based models, few-shot and zero-shot learning, architectural comparisons, and parameter-efficient fine-tuning.

I enhance model performance for diverse languages, with a specialization in improving the understanding and generation of legal summaries. I also work on hallucination mitigation, ensuring summaries are accurate and factually sound. My research encompasses cross-lingual fairness to ensure models perform equitably across languages, as well as the application of techniques like LoRA for efficient model and domain adaptations. I am also exploring tokenless NLP to bypass traditional tokenization and enhance efficiency, as well as interpretable NLP to facilitate transparent decision-making, particularly in legal applications.

In addition to my research, I’ve completed courses on LLMs, Machine and Deep Learning, and NLP on Coursera, edX, and Udacity, further enhancing my technical expertise. I also maintain a Reading Page where I share insights from the latest articles I read and updates on independent courses I’ve audited. This page serves as a resource for anyone interested in the latest trends in technology, natural language processing (NLP), and data science. You can also explore my Courses & Specializations page for more details on the certifications I've completed.

You can find my latest CV here

For more details on my academic background, experience, and skills, please refer to my CV (PDF).